On Thu, 5 Mar 2026 00:33:21 GMT, David Holmes <[email protected]> wrote:

>>> Do you have any more information on this security bug?
>> 
>> I found following bug from mail thread. But I cannot access it due to 
>> security bug.
>> 
>> https://mail.openjdk.org/pipermail/hotspot-runtime-dev/2011-August/002357.html
>> 
>> 4852809: Linux: do not use alternate signal stack
>
>> > Do you have any more information on this security bug?
>> 
>> I found following bug from mail thread. But I cannot access it due to 
>> security bug.
>> 
>> https://mail.openjdk.org/pipermail/hotspot-runtime-dev/2011-August/002357.html
>> 
>> 4852809: Linux: do not use alternate signal stack
> 
> Actually I made this public last year: 
> https://bugs.openjdk.org/browse/JDK-4852809
> 
> There were two main issues:
> 1. Newly created threads inheriting the same alt-stack as parent
> 2. Issues with accessing TLS with Linux fixed-stack implementation
> 
> I don't quite see how #1 could arise, and #2 is definitely not an issue any 
> more.

Hi @dholmes-ora and @YaSuenag !

Sorry for the rebase; something messed up my branch and I had to rebase over 
master to get it clean again. Not sure what happened.

Thank you both for your feedback; I was busy with other stuff, and this moved 
down the pile. I tried to address your concerns. hotspot Tier1 and Tier2 are 
currently running on my machine with the feature enabled.

@dholmes-ora 
> My main concern with this is that it is so hard to test and we will never 
> know to what extent it is getting used. It's usefulness depends entirely on 
> support organizations knowing about it and telling customers to enable this 
> (in production - which they might balk at) to try and better diagnose mystery 
> crashes.

That is true, unfortunately. Whether this is useful enough is a matter of 
taste, but I know I could have used it a bunch of times already, and there are 
an unknown number of JVM-just-vanishes cases that could turn out to be stack 
overflows.

-------------

PR Comment: https://git.openjdk.org/jdk/pull/29559#issuecomment-4118872820

Reply via email to